Retrospective and predictive optimal scheduling of nitrogen liquefier units and the effect of renewable generation

•Development of a binary program to optimally schedule flexible power loads.•Retrospective optimisation to generate a scheduling key performance indicator.•Discovery of renewable generation and spot market correlations for early 2017.•Predictive optimisation of power loads using power pricing foreca...

Full description

Saved in:
Bibliographic Details
Published in:Applied energy 2017-12, Vol.208, p.158-170
Main Authors: Cummings, Thomas, Adamson, Richard, Sugden, Andrew, Willis, Mark J.
Format: Article
Language:English
Subjects:
Binary programming
Predictive optimisation
Price forecasting
Production scheduling
Renewable generation
Retrospective optimisation
Citations: Items that this one cites
Items that cite this one
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:•Development of a binary program to optimally schedule flexible power loads.•Retrospective optimisation to generate a scheduling key performance indicator.•Discovery of renewable generation and spot market correlations for early 2017.•Predictive optimisation of power loads using power pricing forecasts. The construction and application of a multiple nitrogen liquefier unit (NLU) optimal scheduling tool is discussed. Constrained by customer demands and subject to electricity spot market prices over a week-ahead horizon, a retrospective optimiser (RO) determines the minimum scheduling costs. Plant start-up penalties and inter-site optimisation capabilities are incorporated into the optimisation model to emulate realistic operational flexibilities and costs. Using operational data, actual process schedules are compared to the RO results leading to improved process scheduling insights; such as increasing afternoon NLU operation during the spring to utilise lower power pricing caused by high solar generation. The RO is used to output a trackable load management key performance indicator to quantify potential and achieved scheduling improvements. Subsequently, correlations between renewable energy generation and spot market power prices are developed. Forecast pricing is used within a predictive optimiser (PO) to automatically generate an optimal schedule for the week ahead to meet projected customer demands. The RO provides potential hindsight savings of around 11%, and the PO up to 8% (representing significant cost savings for such energy intensive processes).
ISSN:0306-2619
1872-9118
DOI:10.1016/j.apenergy.2017.10.055